[Klug-general] network lags - electrical interference?

[Kevin Groves]

Yes that's the point of twisted pair. Coax has two parallel conductors 
and any simliar parallel conductors with flowing AC current can induce 
current in the other (as Julia says - transformers and yes, regs say 
separation of Band I and Band II class of cables due to induction during 
fault events - a few hunderd Amp no mater how short a time being induced 
into UTP could be quite interesting). Twisted pair appempts to stop this 
parrallel run and generally does it. Don't often have a problem with it.

Regarding the problem I would agree and say hub/switch possible. As 
you've gone Gigabit watch quality of cable and length and any possible 
crap cables. If found you can generally abuse cable length of quality 
quite a bit but one bit of gear that can't cope and there is little you 
can do.

It might be worth checking cable termination as I've found some CAT5 
plugs to look fine but sometimes just won't work in certain sockets. 
Also if it's regular times then sling Wireshark on and see if it's a 
broadcast or repeated retry of packet send/ack.

If you can get hold of one check the cables out with a decent tester 
(not those crappy £10 boxes), I can certainlly recomend the Fluke range 
as I've just recently bought one and it's already earnt it's way. :-)

Kev,


|


On 25/01/10 14:42, Mike Evans wrote:
>
> On 25/01/10 13:50, Alan @ COMM-TECH wrote:
>
>
>> I think therefore that the issue is down to some sort of magnetic or
>> electrical interference, like a faulty lighting circuit or some leery
>> telephone cable crossing
>>
> Twisted pair is remarkably immune to interference.  Back in the early
> days of local area networking for PCs when I worked for Novell we were
> developing some network cards and comparing performance over ethernet
> cable (which used to be coax) and twisted pair.  The coax would quite
> often loose packets when the soldering iron was placed close to the reel
> of cable we were using as our test.  (Those soldering irons that have a
> thermal cut-out built in.) Twisted pair took all manner of abuse - we
> could even run an electric drill with worn brushes next to it.
>
> The fact that your drop-out runs for variable lengths of time does
> indicate that there is probably a co-incidence with some sort of human
> activity if only you can work out what it is.
>
> _______________________________________________
> Kent mailing list
> Kent at mailman.lug.org.uk
> https://mailman.lug.org.uk/mailman/listinfo/kent
>
>